A fundamental challenge in research with extreme ultraviolet radiation is the lack of optics with sufficient reflectance. Apart from gracing incidence mirrors there exist no optics in the extreme ultraviolet, which offer reflectances comparable to those of silver- coated mirrors in the visible or gold-coated mirrors in the infrared. Therefore progress in experiments with attosecond laser pulses has always been compromised by the limited usability of gracing incidence optics [28,69] or depended strongly on the development of suitable XUV multilayer optics [6]. At normal incidence the peak reflectance of state-of- the-art multilayer mirrors never exceeds 15%, if a spectral bandwidth of of more than 10 eV (FWHM) has to be reflected. The situation becomes even more complicated if very broadband reflectances of more than 20 eV are required. Such broadband XUV multi- layer mirrors suffer from a peak reflectance of less than 5% [7]. Moreover, XUV multilayer mirrors suitable for attosecond laser pulses cannot be designed in any arbitrary spectral range, since the layers are made of different materials in order to modulate the imaginary part𝐼𝑚(𝑛) of the refractive index𝑛. A standard material for XUV multilayer mirrors is silicon, which shows a strong absorption edge at 100 eV. Such an absorption edge causes strong phase effects preventing the preservation of the shortest possible pulse duration of the incident XUV pulses. In summary, even nowadays normal incidence XUV multilayer optics reveal a low reflectance, which is decreasing with an increasing reflected spectral bandwidth. On top of this, the reflected spectral bandwidth is limited and can only be designed in a few spectral windows of the XUV specrum.
All the difficulties with normal incidence XUV optics justify - besides other arguments, which have already been highlighted in chapter4.1- the invention of the AS2-beamline, since the angle of incidence of the XUV radiation on the XUV optic in the delay chamber in Fig. 19 on page 41 is 45∘ instead of normal incidence. On top of that, it turned out that multilayer XUV mirrors can be replaced by simple metal coated mirrors if several of these metal coated mirrors are sequentially used in order to increase the angle of incidence on each XUV-mirror. For example, three mirrors with an angle of incidence of 75∘ result in a total deflection of the XUV beam by 90∘ as required in the setup of the AS2 beamline.
This concept has been realized as shown in Fig. 49:
Fig. 49: left: design of a mirror mount, which deflects the XUV beam by 90∘ after 3 subsequent reflections at 75∘ angle of incidence, right: mirror mount for rhodium-coated XUV mirrors in the delay chamber of the AS2 beamline. The half-inch mirror on the top is part of the reference interferometer for the active stabilization of the Mach-Zehnder interferometer.
The spectral reflectance of metal coatings can easily be derived from the complex re- fractive index [29,83]. In Fig. 50 the total reflectance after three reflections on metal coated mirrors at an angle of incidence of 75∘ is shown for ruthenium, rhodium and palladium [29]. All these coatings are commercially available and stable in air. In ad- dition the reflectance curve of a normal incidence multilayer mirror is shown supporting a spectral bandwidth of about 25 eV (FWHM), which has been used for the generation of the shortest XUV pulses so far with a pulse duration of only 80 attoseconds [7,81]. Apparently the application of metal coated XUV mirrors in the AS2 beamline allows an increase of the XUV flux on target by more than 2 orders of magnitude and an increase of the supported spectral bandwidth (FWHM) by a factor of 2. This invention opens the door to the generation of XUV pulses on target with pulse durations below 80 at- toseconds at 100 times higher pulse energy. Another advantage of metal coated mirrors is that they preserve the phase of the XUV electric field, which is an important point, since many XUV multilayer mirror designs exhibit a useful spectral reflectance curve but cannot be used for the reflection of attosecond laser pulses due to disadvantageous phase effects.
Fig. 50: Comparison of the total reflectance after 3 subsequent reflections on 3 metal coated mirrors at 75∘ angle of incidence and the reflectance at normal incidence of one multilayer mirror, which has been used for the generation of single XUV pulses with a pulse duration of 80 attoseconds [7].
4.5.2 Streaking measurements
A first streaking spectrogram with these novel metal coated XUV mirrors is shown in Fig. 51. In the past, all photoelectron spectra have been measured with a commercial electron time-of-flight spectrometer, which increases the acceptance angle of the emmitted photoelectrons by means of an electrostatic lens. Thus, it increases the photoelectron count rate by several orders of magnitude, which was a condition for the feasability of many attosecond experiments in the past due to the low XUV flux and absorption cross sections in the XUV. This lens supports only a limited spectral bandwidth of about 40 eV at a central energy of 100 eV. In case of 30 eV broad photoelectron spectra, which are streaked by additional 15 eV as shown in Fig. 51, this electrostatic lens fails since its supported spectral bandwidth is too small. Fortunately, by means of rhodium-coated XUV mirrors the XUV flux could be sufficiently increased so that streaking measurements at tolerable integration times per delay step are still doable. Rhodium as coating material in combination with a molybdenum filter were selected since rhodium shows the highest reflectance in the spectral range of the generated XUV continua between 80 eV and 120 eV. The shown streaking scan reveals a slight positive chirp of the XUV pulses. This positive chirp originates from the HHG with short trajectories as described in chapter 3.1.1and is not completely compensated by the negative chirp introduced by the 150 nm thick molybdenum filter, which has been used as a high pass filter that transmits the continuous cut-off of the high harmonic spectrum (see Fig. 6 on page 18). The residual positive chirp of the XUV pulses could be compensated by a thicker metal filter, which
should result in a shorter attosecond pulse duration.
Fig. 51: Streaking scan with 150 nm Mo as spectral filter and rhodium mirrors. The integration time per delay step was 20 sec, the delay stepsize was 50 attoseconds.
Fig. 52 shows the temporal XUV intensity profile, which was retrieved by a FROG analysis of the streaking scan in Fig. 51. The XUV pulse duration is 77 attoseconds (FWHM).
Fig. 52: Temporal XUV intensity profile retrieved from the streaking scan shown in Fig. 51.
5 Capturing electron dynamics
5.1 Time-resolved measurement of electron tunneling